CN106324708B - Digitizing solution, the device of rainfall record drawing - Google Patents
Digitizing solution, the device of rainfall record drawing Download PDFInfo
- Publication number
- CN106324708B CN106324708B CN201510355671.4A CN201510355671A CN106324708B CN 106324708 B CN106324708 B CN 106324708B CN 201510355671 A CN201510355671 A CN 201510355671A CN 106324708 B CN106324708 B CN 106324708B
- Authority
- CN
- China
- Prior art keywords
- rainfall
- image
- line
- binaryzation
- point
- 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.)
- Expired - Fee Related
Links
Abstract
The invention discloses a kind of digitizing solutions of rainfall record drawing.It includes the following steps:Extract grid image, the rainfall line image of binaryzation respectively from the digital picture of rainfall record drawing;To grid image, rainfall line image into line tilt correction;The boundary of net region is determined from the grid image after slant correction using the method for ranks projection, and sets the grid image after slant correction, the boundary exterior domain in rainfall line image to binaryzation background;Set the straight line in rainfall line image to binaryzation background;Rainfall line image is corresponded into grid image, the maximum of points of each peak point and rightmost side peak point right side extended line on rainfall line is found out, finally sums to the rainfall value corresponding to the above each point, obtain rainfall.The invention also discloses a kind of digitalizers of rainfall record drawing.The present invention can quickly and accurately automatically extract out the rainfall product data recorded from rainfall record line drawing, be analyzed convenient for subsequent data query.
Description
Technical field
The present invention relates to a kind of digitizing solution, the devices of rainfall record drawing, are remembered using Computer Automatic Recognition rainfall
The rainfall data that record drawing is recorded belongs to what hydrometeorology, digitized image processing, image recognition and database intersected
Technical field.
Background technology
All it is by specific in fields such as water conservancy, meteorology, biomedicine, petroleum geologies before computer is not universal
The valid data information of observation is manually plotted on papery engineering drawing by instrument.Data on these papery hydrological datas
It is the precious historical summary of comparison, record flow, water level, precipitation, evaporation capacity etc., wherein many data can be by through common
It further analyzes and handles, the utilization to these effective informations and the shared reason that science will be provided for China's hydraulic engineering construction
By with foundation, it will greatly promote the development of China's water conservancy.
However this kind of hydrology papery data be not it is complete preserve it is complete, with the time, can by storage condition, temperature,
The influences such as environment cause such as papery to be ground away breakage, thin out fuzzy, the papery data writing infiltration dirt of time aging too long, writing
Dye, these can all affect the safety and value of historical data.Therefore this kind of papery data is digitized, by paper
It is necessary that important data in matter hydrological data, which are transferred in computer and carry out classification storage and processing,.
Since hydrology papery data type is various and data are diversified, these data are carried out with the mistake for being saved in computer
Cheng Zhong selects representative node and data usually based on artificial, and the method for specifying interpolation or curve matching calculates
The data such as water level information sometime.But tracing pattern is different, fitting or interpolation can not be general.Another way is logical
It crosses artificial logging data and establishes large database, since the various data volume of data type is big, workload is huge and error-prone, number
According to need by stringent examination with check and could be put in storage, and need to provide a variety of different retrieval modes, and do not guarantee that
The data of needs can be accurately found in the shortest time.
In miscellaneous hydrology papery data, rainfall record drawing has its particularity again.Every Pluviogram all can
The rainfall of record 1 day or more, the corresponding recording curve in day without rainfall is a horizontal linear.And it is corresponding with rainfall
Curve record on the day of 8 when by second day 8 when rainfall line, it since a certain lower scale, if this 24 hours
Interior rainfall has been more than measuring cup (full measuring cup be 10mm), just pours out the rainwater of measuring cup and continues to measure, thus curve can be formed it is several
A peak, peak value is between 9-10mm.Rainfall in last time measuring cup goes up always, but as long as measuring cup is less than, then homologous thread
On the less than call wire of measuring cup.The rainfall for calculating this day needs the sum for acquiring each peak value and the less than record value of measuring cup, finally
Obtain rainfall.For rainfall record paper, there is following feature:Due to needing record rainfall in 24 hours, note
Record paper web width is larger, and the image pixel that camera is shot is more, easy tos produce the even phenomenon of uneven illumination, shows same root curve
Different piece color is inconsistent;Due to rainfall complexity, in continuous rain in torrents, peak shows that rainfall line is steep than comparatively dense
Drop skyrockets;And it, both may be parallel to each other between call wire, it is also possible to mutually due to having recorded more days rainfall datas on a piece of paper
Staggeredly, traditional images processing method is not easy the change in process of every line of trace analysis.
Invention content
The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and to provide a kind of numbers of rainfall record drawing
Word method can quickly and accurately automatically extract out the rainfall product data recorded, just from rainfall record line drawing
It is analyzed in subsequent data query.
The present invention specifically uses following technical scheme to solve above-mentioned technical problem:
The digitizing solution of rainfall record drawing, the rainfall number recorded using Computer Automatic Recognition rainfall record drawing
According to including the following steps:
Step A, grid image, the rainfall line chart of binaryzation are extracted respectively from the digital picture of rainfall record drawing
Picture;
Step B, to grid image, rainfall line image into line tilt correction;
Step C, the boundary of net region is determined from the grid image after slant correction using the method for ranks projection, and
Set the grid image after slant correction, the boundary exterior domain in rainfall line image to binaryzation background;
Step D, the straight line in rainfall line image is set to binaryzation background;
Step E, rainfall line image is corresponded into grid image, finds out each peak point and rightmost side peak value on rainfall line
The maximum of points of point right side extended line, finally sums to the rainfall value corresponding to the above each point, obtains rainfall.
Preferably, the rainfall line chart of binaryzation is extracted from the digital picture of rainfall record drawing using following methods
Picture:
Step 1 carries out gaussian pyramid resampling processing to the digital picture of rainfall record drawing, obtains the digitized map
The low-resolution image of picture;
Step 2 carries out preliminary binary conversion treatment to the low-resolution image, obtains the rainfall line chart of preliminary binaryzation
Picture;
Step 3 divides the rainfall line image of the digital picture of rainfall record drawing, low-resolution image, preliminary binaryzation
It is not divided into N number of subgraph from left to right according to same dividing mode, N is the integer more than 1;
Step 4, the rainfall line image for finding out preliminary binaryzation i-th of subgraph in each target pixel points in low resolution
Corresponding pixel points in i-th of subgraph of rate image, i=1,2 ..., N, and calculate these in CIE Lab color spaces
Maximum space distance dTh between pixel and its average point (L_avg, a_avg, b_avg);
Step 5, traverse rainfall record drawing digital picture i-th of subgraph in all pixels point, will meet with
The pixel of lower condition is set as binaryzation target, and rest of pixels point is set as binaryzation background:In CIE Lab color spaces
The distance between average point (L_avg, a_avg, b_avg) is less than dTh, and the H components in HSI color spaces are more than hTh,
HTh is the default H components threshold value more than or equal to 180;To obtain the rainfall line image of final binaryzation.
Further, it is described to grid image, rainfall line image into line tilt correction, it is specific as follows:From preliminary binaryzation
Rainfall line image in find out longest straight line, the angle of inclination of the straight line is then detected using Hough transformation method, finally
Rotation correction is carried out to grid image, rainfall line image according to the angle of inclination detected.
Preferably, preliminary binary conversion treatment is carried out to low-resolution image using improved Niblack algorithms, specifically such as
Under:First by the low-resolution image gray processing;Then the binary-state threshold of each pixel is determined using following formula:
Wherein T (x, y) is the binary-state threshold of point (x, y), and m (x, y) is that point in size template is preset around point (x, y)
Gray average;S (x, y) indicates the standard deviation of pixel gray value in template;The value range of parameter k is [0.4,0.6], and R is
128;
If the gray value at point (x, y) is less than T (x, y), then it is set and is otherwise set as target for background.
Following technical scheme can also be obtained according to identical invention thinking:
The digitalizer of rainfall record drawing, including:
Grid image extraction unit, the grid chart for extracting binaryzation from the digital picture of rainfall record drawing
Picture;Rainfall line image extraction unit, the rainfall line image for extracting binaryzation from the digital picture of rainfall record drawing;
Tilt corrector unit, the grid chart for being exported to grid image extraction unit and rainfall line image extraction unit
As, rainfall line image is into line tilt correction;
Grid clipping unit, for determining grid regions from the grid image after slant correction using the method for ranks projection
The boundary in domain, and set the grid image after slant correction, the boundary exterior domain in rainfall line image to binaryzation background;
Line removal unit, for setting the straight line in rainfall line image to binaryzation background;
Rainfall computing unit finds out each peak point on rainfall line for rainfall line image to be corresponded to grid image
And the maximum of points of extended line is dropped finally to the rainfall value summation corresponding to the above each point on the right side of the peak point of the rightmost side
Rainfall.
Preferably, the rainfall line image extraction unit specifically includes:
Gaussian pyramid resampling module carries out gaussian pyramid resampling for the digital picture to rainfall record drawing
Processing, obtains the low-resolution image of the digital picture;
Preliminary binary processing module obtains preliminary for carrying out preliminary binary conversion treatment to the low-resolution image
The rainfall line image of binaryzation;
Image division module is used for the rain of the digital picture of rainfall record drawing, low-resolution image, preliminary binaryzation
Amount line image is divided into N number of subgraph from left to right according to same dividing mode respectively, and N is the integer more than 1;
Threshold value acquisition module, each target picture in i-th of subgraph of the rainfall line image for finding out preliminary binaryzation
Corresponding pixel points of the vegetarian refreshments in i-th of subgraph of low-resolution image, i=1,2 ..., N, and in CIE Lab color spaces
In calculate maximum space distance dTh between these pixels and its average point (L_avg, a_avg, b_avg);
Rainfall line image output module, the institute in i-th of subgraph of the digital picture for traversing rainfall record drawing
There is pixel, sets the pixel for meeting the following conditions to binaryzation target, rest of pixels point is set as binaryzation background:
It is less than dTh with the distance between average point (L_avg, a_avg, b_avg) in CIE Lab color spaces, and in HSI color spaces
In H components be more than hTh, hTh is default H components threshold value more than or equal to 180;To obtain the rainfall of final binaryzation
Line image.
Further, the tilt corrector unit carries grid image extraction unit and rainfall line image using following methods
Take the grid image that unit exported, rainfall line image into line tilt correction:It is utilized from the rainfall line image of preliminary binaryzation
Hough transformation method detects the angle of inclination of longest straight line, finally according to the angle of inclination detected to grid image, rainfall
Line image carries out rotation correction.
Preferably, the preliminary binary processing module carries out low-resolution image using improved Niblack algorithms
Preliminary binary conversion treatment, it is specific as follows:First by the low-resolution image gray processing;Then each picture is determined using following formula
The binary-state threshold of vegetarian refreshments:
Wherein T (x, y) is the binary-state threshold of point (x, y), and m (x, y) is that point in size template is preset around point (x, y)
Gray average;S (x, y) indicates the standard deviation of pixel gray value in template;The value range of parameter k is [0.4,0.6], and R is
128;
If the gray value at point (x, y) is less than T (x, y), then it is set and is otherwise set as target for background.
Further, which further includes the image acquisition unit of the digital picture for obtaining rainfall record drawing, with
And the data storage cell for storing the exported rainfall product data of rainfall computing unit.
Compared with prior art, the invention has the advantages that:
The present invention can quickly and accurately automatically extract out the rainfall number recorded from rainfall record line drawing
According to, convenient for subsequent data query analyze, the digitized efficiency of rainfall data greatly improved, reduce manpower and time at
This;
The present invention obtains low-resolution image when carrying out the extraction of rainfall line image, first with gaussian pyramid model, so
Preliminary binary conversion treatment is carried out to low-resolution image afterwards, finally combines rainfall record line drawing CIE Lab color spaces
With HSI color space features, preliminary binary conversion treatment result is advanced optimized, calculation amount is effectively reduced, improved
Extraction rate, and the result obtained is more accurate;
The present invention is with the longest straight line in the rainfall line image of preliminary binaryzation as a reference to carrying out Hough transformation inclination
Correction, algorithm is simpler, and accuracy is more preferable.
Description of the drawings
Fig. 1 is practical rainfall record drawing;
Fig. 2 is the flow diagram of digitizing solution of the present invention in specific implementation mode;
Fig. 3 is the rainfall line image of preliminary binaryzation;
Fig. 4 is the rainfall line image after double optimization;
Fig. 5 is the grid image of the binaryzation extracted;
Fig. 6 is the rainfall line image after slant correction;
Fig. 7 is the grid image after slant correction;
Fig. 8 is the rainfall line image after cutting;
Fig. 9 is the grid image after cutting;
Figure 10 is the rainfall line image removed after straight line.
Specific implementation mode
Technical scheme of the present invention is described in detail below in conjunction with the accompanying drawings:
Fig. 1 shows a practical rainfall record drawing, as shown, every Pluviogram can all record 1 day or more drop
Rainfall, the corresponding recording curve in day without rainfall are a horizontal linears.And it is corresponding with 8 on the day of the curve record of rainfall
When by second day 8 when rainfall line, it since a certain lower scale, if in this 24 hours rainfall be more than amount
Cup, the water of measuring cup is just poured out and continues to measure, therefore curve can form several peaks, peak value is between 9-10mm.For the last time
Rainfall in measuring cup goes up always, but as long as measuring cup is less than, then the less than call wire of measuring cup on homologous thread.Calculate the drop of this day
Rainfall needs the sum for acquiring each peak value and the less than record value of measuring cup, finally obtains rainfall.
The invention firstly uses digital image processing techniques to extract grid image from the digital picture of rainfall record drawing
With rainfall line image, then the image extracted is corrected, the processing such as denoising, then according to rainfall line peak value and extension
Line extreme value calculates rainfall value.Fig. 2 shows the basic procedure of the method for the present invention, specifically includes following steps:
Step A, grid image, the rainfall line chart of binaryzation are extracted respectively from the digital picture of rainfall record drawing
Picture.
1, rainfall line drawing:
Since rainfall record drawing is bigger, the Pluviogram resolution ratio shot with digital camera is all very high, generally reaches
The pixel scale of 4000*3000.As directly handled artwork (being usually the RGB bitmaps of standard), it is clear that it is computationally intensive, it carries
It takes speed slow, seriously affects rainfall record digitalized drawing efficiency.To solve this problem, the present invention when extracting rainfall line simultaneously
Rainfall record artwork I is not utilized directly0It is handled, but relatively low resolution is obtained by this pyramid model of structural map image height
Rate image carries out gaussian pyramid resampling processing, then to low-resolution image to the digital picture of rainfall record drawing
It is handled.Pyramid series can be determined according to artwork resolution ratio used by specific, for example, for 4000*3000 scales
Three-level gaussian pyramid model can be used in artwork, to obtain low-resolution image I2.Image gaussian pyramid is existing maturation
Technology, to save space, details are not described herein again, and detail can refer to document (Rafale C Gonzalez, Richard E
Woods writes Digital Image Processing [M] Ruan Qiu fine jades, and Ruan Yuzhi etc. translates the Beijing:Electronic Industry Press, 2004.).
To low-resolution image I2Preliminary binary conversion treatment is carried out, the rainfall line image of preliminary binaryzation is obtained.It can adopt
Preliminary binarization segmentation is carried out with existing Niblack algorithms, gray scale expected value method, Otsu methods etc., present invention preferably employs
Improved Niblack algorithms (J.Sauvola, M.Pietikainen.Adaptive document image
Binarization.Pattern Recognition, 33 (2000), pp.225-236), it is specific as follows:
First by low-resolution image I2Gray processing is carried out, then carries out binaryzation using following improved Niblack formula
Processing:
Wherein T (x, y) is the threshold value of point (x, y), m (x, y) be preset around point (x, y) size template (usually 9*9~
Chosen between 15*15, preferably 11*11) in put gray average;S (x, y) indicates the standard deviation of pixel gray value in template;
The value range of parameter k be [0.4,0.6], preferably 0.5;Since we color image into 8 gray level images, so
R can be set as 128.Gray value at point (x, y) is less than T (x, y), then the value of its bianry image is set as 0, is otherwise set as 255, from
And obtain bianry image I2 B, thus can extract out the major part of rainfall curve.Fig. 3 is shown using improvement Niblack algorithms institute
Obtain an example of preliminary binaryzation rainfall line image.
The rainfall curve tentatively extracted is due to very big by background interference, so obtained bianry image contains many noises
And redundant information, while if rainfall curve is thinner, white pixel point large percentage in a template so that threshold value is inclined
Greatly, thus obtained curve breakpoint is more, therefore the present invention advanced optimizes preliminary bianry image, is extracted more with this
Complete rainfall line image.Optimization method of the present invention is specific as follows:
1. since rainfall record drawing length is larger, uneven illumination is easy tod produce when shooting, in order to reduce illumination to rain
The influence for measuring curve, first by image I0、I2、I2 BMultiple subgraphs from left to right are divided into according to same dividing mode,
Image is divided into the part of left, center, right three in the present embodiment;
2. by taking left one third as an example, by image I2Three components normalization in CIE Lab color spaces;
3. bianry image is mapped to CIE Lab coloured images I2In, statistics is acquired with the black color dots in bianry image (i.e.
The point on rainfall line tentatively obtained) corresponding to point L, a, the mean value L_avg, a_avg, b_avg of b components;
4. calculating CIE Lab coloured images I2Middle each point to (L_avg, a_avg, b_avg) Euclidean distance dis,
Determination can extract the threshold value dTh of curve, and the present invention is using the point on the rainfall line tentatively obtained to (L_avg, a_avg, b_
Avg maximum distance) is as dTh;
5. being found through a large amount of analysis and observations, the rainfall curve H component values in original image HSI color model are usually both greater than
180, therefore the present invention is further using the value or somewhat larger value as the H component threshold value hTh of HSI color spaces, it is of the invention
It is preferred that the value of hTh is 200;
6. traversing original image I0In all the points, set the pixel for meeting the following conditions to binaryzation target,
Afterimage vegetarian refreshments is set as binaryzation background:In CIE Lab color spaces between average point (L_avg, a_avg, b_avg)
Euclidean distance is less than dTh, and the H components in HSI color spaces are more than hTh;That is, if distance dis<DTh and H>
The value of the pixel is then set as 0, is otherwise set as 255 by hTh;
7. repeating one third in realizing according to the method described above, right one third part obtains the rainfall song of double optimization
Line.The preliminary rainfall line image of Fig. 3 is using as shown in Figure 4 after above method optimization.
2, grid image extracts
All it is in artwork since the extraction of grid lines needs to extract all lateral graduation marks and longitudinal graduation mark
I0On handled, present invention preferably employs Chinese invention patent application (Publication No. 103971367A, the applyings date 2014/
4/28) the grid image extracting method disclosed in, is as follows:
(1) horizontal, vertical direction gradient distribution is carried out to the image of gray processing using Sobel gradient operators, obtains two
Gradient map on direction is to obtain the marginal information of target lines;
(2) in gradient map both horizontally and vertically, be respectively processed (for this sentences horizontal direction, Vertical Square
To processing mode it is identical):
1. in order to reduce the influence of uneven illumination, three pieces of left, center, right is divided into horizontal gradient figure, in corresponding fritter, in conjunction with
The curve position of curve extraction, weeds out the influence of the Grad of homologous thread, the ladder of grid lines is belonged in calculation block
Spend mean value Th;
2. carrying out binaryzation to horizontal gradient image using gradient mean value Th, on horizontal gradient gradient distribution direction, separately
The horizontal gradient of the pixel condition bigger than vertical gradient is added so that horizontal gradientization is more accurate, according to this binaryzation knot
Fruit tentatively obtains the location information for belonging to orange-yellow lines;
3. according to the acquisition yellow line location information in 2., obtains the pixel in horizontal direction on corresponding position and exist
A the and b component sizes of CIE Lab spaces, and count a, b mean value of these location points
4. utilizing what is obtained in 3.If a, b component of pixel meetThen be target, otherwise for
Background is split original image.
5. obtaining horizontal direction binary picture;Same method obtains the binary picture of vertical direction.
(3) binary picture both horizontally and vertically is carried out or operation merges, obtain whole image binary picture;
(4) corrosion of horizontal direction 1 × 5 is carried out respectively for the binary picture obtained in (4) and preserved as a result, and vertical
5 × 1 corrosion preserves as a result, getting rid of more discrete points on direction;
(5) binary picture obtained in horizontal and vertical directions is merged, obtains final grid image.Figure
5 show the grid image extracted.
Step B, to grid image, rainfall line image into line tilt correction.
In the image process for obtaining rainfall record drawing, since its breadth is larger, it frequently can lead to image and difference occur
The inclination of degree, it is therefore necessary to image into line tilt correction.Existing slant correction is typically the side using Hough transformation
Method acquires the angle of picture deflection, then rotation correction.In view of that can have slightly long straight line (no rainfall in rainfall line image
Rainfall curve), be convenient for identifying processing, therefore the present invention finds out longest straight line from the rainfall line image of preliminary binaryzation,
Then Hough (Hough) transform method is utilized to detect the angle of inclination of the straight line, finally according to the angle of inclination pair detected
Grid image, rainfall line image carry out rotation correction;Its detailed process is as follows:
(1) in the bianry image I of smaller scale2 BUpper traversal black pixel point (i, j) utilizes | i*cos α+j*sin α | it asks
It obtains (i, j) and arrives the distance of origin, and be saved into accumulator array;
(2) maximum value of accumulator array, i.e. bianry image I are found2 BIn longest straight line, acquire corresponding deflection angle α;
(3) artwork is rotated clockwise according to angle [alpha], conversion formula:I1=i*cos α+j*sin α, j1=i*cos α-j*
Sin α, wherein (i, j) and (i1, j1) is respectively the pixel coordinate of artwork and rotated image, after obtaining angle tilt correction
Rainfall line image and grid image, it is as shown in Figure 6, Figure 7 respectively.
Step C, the boundary of net region is determined from the grid image after slant correction using the method for ranks projection, and
Set the grid image after slant correction, the boundary exterior domain in rainfall line image to binaryzation background.
Postrotational rainfall line image and grid image contain many record drawings with the information of outer portion, it is therefore desirable to
Meshing is cut out to come, avoids the interference of other information.The specific method is as follows for it:
(1) grid image row, column is projected, counts the black pixel point number of each row (column);
(2) according to row (column) project statistics given threshold, herein row threshold value take 200, row threshold value takes 400;
(3) grid lines boundary up and down is found according to the ranks threshold value of setting;
(4) part outside net boundary is all set to white, the image after being cut;Rainfall line chart after cutting
Picture, grid image difference are as shown in Figure 8, Figure 9.
Step D, the straight line in rainfall line image is set to binaryzation background.
All include several days rainfall datas in usual rainfall record figure, and be wherein a few days ago not rainfall,
Rainfall record is exactly a straight line, therefore can erase the straight line in rainfall line image, to avoid to subsequently there is rainfall
Rainfall record identification generates interference.Following method specifically can be used:
(1) 450 row to rainfall line image in the range of grid since grid left end do floor projection, count black
The number of pixel;
(2) given threshold Th, when the floor projection statistical value of certain row then assert straight line on this journey more than Th, and first
The secondary line number occurred more than Th is left the record of rainy day;
(3) gray value of all pixels point is set as 255 in 20 rows above and below line number of the statistical value more than Th, is erased
Rainfall line image after straight line, as shown in Figure 10.
Step E, rainfall line image is corresponded into grid image, finds out each peak point and rightmost side peak value on rainfall line
The maximum of points of point right side extended line, finally sums to the rainfall value corresponding to the above each point, obtains rainfall.
Utilize the obtained binaryzation rainfall line image of each step, grid image above, you can seek out rainfall value.Rainfall
Value includes two parts:When the record value corresponding to each peak point of rainfall curve, second is that less than call wire (the i.e. rightmost side of measuring cup
The maximum of points of extended line on the right side of peak point) value.The method for seeking rainfall value is specific as follows:
(1) width for calculating each scale of grid lines is high
1. doing ranks projection to the grid lines upper left corner regions 300*300, each row (column) black pixel point number is counted;
2. given threshold Th finds the transverse direction of grid and vertical graduation mark;
3. calculating the average value of horizontal (perpendicular) to pixel between graduation mark, it is denoted as height (width) degree of grid.
(2) curve peak is found
1. the section of scale 9-10 is divided into several fritters by mesh width, of each fritter inner curve point is counted respectively
Number;
2. given threshold Th, when step assert the peak in the region there are curve when 1. statistical value is more than Th;
3. finding peak in the 2. grid block comprising curve peak that step obtains, and judge 25 pixels on the right of the point
Inside whether there is the point also higher than it, if it is not, assert that the point is a peak point, its coordinate is preserved;
(3) peak of the less than call wire of measuring cup is found
The region turned right to the last one peak point is dealt with, and finds the highest of the less than call wire of measuring cup in this panel region
Point, the number by counting black pixel point in mesh width of the point or so judge that this point is point on curve or isolated
Noise.If it is the point on curve, then the point is preserved.
(4) rainfall value is calculated
1. in grid image, extreme point (x, y) into next grid altitude range, 100 pixels each to y or so
Floor projection is done in region, counts the number of often row black pixel point, and the minimum abscissa more than 140 is regarded as under (x, y)
Graduation mark;
2. it is 1. similar with step, from (x, y) up until grid top in the range of, 100 pixels each to y or so
Region do floor projection, obtain each graduation mark of the upper end (x, y);
3. taking the graduation mark up and down near (x, y), (x, y) is calculated between the height h and upper and lower graduation mark of lower graduation mark
Height H, calculate h/H.
4. counting the item number i of the upper end (x, y) graduation mark, value (10-i)+h/H corresponding to (x, y) is calculated;
5. calculating the value of each extreme point using above-mentioned steps, and the peak on the less than record line of measuring cup is found out,
In addition several peak values of front, finally obtain rainfall.
The digitalizer of rainfall record drawing of the present invention can be obtained according to the method for the present invention, including:
Grid image extraction unit, the grid chart for extracting binaryzation from the digital picture of rainfall record drawing
Picture;Rainfall line image extraction unit, the rainfall line image for extracting binaryzation from the digital picture of rainfall record drawing;
Tilt corrector unit, the grid chart for being exported to grid image extraction unit and rainfall line image extraction unit
As, rainfall line image is into line tilt correction;
Grid clipping unit, for determining grid regions from the grid image after slant correction using the method for ranks projection
The boundary in domain, and set the grid image after slant correction, the boundary exterior domain in rainfall line image to binaryzation background;
Line removal unit, for setting the straight line in rainfall line image to binaryzation background;
Rainfall computing unit finds out each peak point on rainfall line for rainfall line image to be corresponded to grid image
And the maximum of points of extended line is dropped finally to the rainfall value summation corresponding to the above each point on the right side of the peak point of the rightmost side
Rainfall.
The pre- rainfall record figure for first passing through the acquisition of the equipment such as scanner, digital camera can be used in the digitalizer of the present invention
The image of paper is as input, and the rainfall data that rainfall computing unit is obtained is stored in external data base, so as to subsequent point
Analysis inquiry;Image acquisition equipment and/or data storage device further can also be integrated by the present invention, realize full-automatic
Rainfall data extraction.
The experimental results showed that the present invention can automatically carry out number to the rainfall record drawing using digital camera shooting
Change analysis, has outstanding recognition capability, the value accuracy of the rainfall of acquisition also very high in common rainfall record figure.
Claims (8)
1. the digitizing solution of rainfall record drawing, the rainfall number recorded using Computer Automatic Recognition rainfall record drawing
According to, which is characterized in that include the following steps:
Step A, grid image, the rainfall line image of binaryzation are extracted respectively from the digital picture of rainfall record drawing;Make
The rainfall line image of binaryzation is extracted from the digital picture of rainfall record drawing using the following method:
Step 1 carries out gaussian pyramid resampling processing to the digital picture of rainfall record drawing, obtains the digital picture
Low-resolution image;
Step 2 carries out preliminary binary conversion treatment to the low-resolution image, obtains the rainfall line image of preliminary binaryzation;
Step 3 presses the rainfall line image of the digital picture of rainfall record drawing, low-resolution image, preliminary binaryzation respectively
It is divided into N number of subgraph from left to right according to same dividing mode, N is the integer more than 1;
Step 4, the rainfall line image for finding out preliminary binaryzation i-th of subgraph in each target pixel points in low resolution figure
Corresponding pixel points in i-th of subgraph of picture, i=1,2 ..., N, and calculate these pixels in CIE Lab color spaces
Maximum space distance dTh between point and its average point (L_avg, a_avg, b_avg);
Step 5, traverse rainfall record drawing digital picture i-th of subgraph in all pixels point, following item will be met
The pixel of part is set as binaryzation target, and rest of pixels point is set as binaryzation background:In CIE Lab color spaces and
It is worth the distance between point (L_avg, a_avg, b_avg) and is less than dTh, and the H components in HSI color spaces is more than hTh, hTh
For the default H components threshold value more than or equal to 180;To obtain the rainfall line image of final binaryzation;
Step B, to grid image, rainfall line image into line tilt correction;
Step C, the boundary of net region is determined from the grid image after slant correction using the method for ranks projection, and will be inclined
Tiltedly the grid image after correction, the boundary exterior domain in rainfall line image are set as binaryzation background;
Step D, the straight line in rainfall line image is set to binaryzation background;
Step E, rainfall line image is corresponded into grid image, it is right finds out each peak point and rightmost side peak point on rainfall line
The maximum of points of side extended line finally sums to the rainfall value corresponding to the above each point, obtains rainfall.
2. the digitizing solution of rainfall record drawing as described in claim 1, which is characterized in that the value of N is 3.
3. the digitizing solution of rainfall record drawing as described in claim 1, which is characterized in that described to grid image, rainfall
Line image is specific as follows into line tilt correction:It is detected using Hough transformation method from the rainfall line image of preliminary binaryzation
The angle of inclination of longest straight line finally carries out rotation correction according to the angle of inclination detected to grid image, rainfall line image.
4. the digitizing solution of rainfall record drawing as described in claim 1, which is characterized in that calculated using improved Niblack
Method carries out preliminary binary conversion treatment to low-resolution image, specific as follows:First by the low-resolution image gray processing;Then
The binary-state threshold of each pixel is determined using following formula:
Wherein T (x, y) is the binary-state threshold of point (x, y), and m (x, y) is that the gray scale put in size template is preset around point (x, y)
Mean value;S (x, y) indicates the standard deviation of pixel gray value in template;The value range of parameter k is [0.4,0.6], R 128;
If the gray value at point (x, y) is less than T (x, y), then it is set and is otherwise set as target for background.
5. the digitalizer of rainfall record drawing, which is characterized in that including:
Grid image extraction unit, the grid image for extracting binaryzation from the digital picture of rainfall record drawing;
Rainfall line image extraction unit, the rainfall line chart for extracting binaryzation from the digital picture of rainfall record drawing
Picture;
Tilt corrector unit, grid image for being exported to grid image extraction unit and rainfall line image extraction unit,
Rainfall line image is into line tilt correction;
Grid clipping unit, for determining net region from the grid image after slant correction using the method for ranks projection
Boundary, and set the grid image after slant correction, the boundary exterior domain in rainfall line image to binaryzation background;
Line removal unit, for setting the straight line in rainfall line image to binaryzation background;
Rainfall computing unit, for rainfall line image to be corresponded to grid image, find out on rainfall line each peak point and
The maximum of points of extended line on the right side of the peak point of the rightmost side, finally sums to the rainfall value corresponding to the above each point, obtains rainfall;
The rainfall line image extraction unit specifically includes:
Gaussian pyramid resampling module is carried out for the digital picture to rainfall record drawing at gaussian pyramid resampling
Reason, obtains the low-resolution image of the digital picture;
Preliminary binary processing module obtains preliminary two-value for carrying out preliminary binary conversion treatment to the low-resolution image
The rainfall line image of change;
Image division module is used for the rainfall line of the digital picture of rainfall record drawing, low-resolution image, preliminary binaryzation
Image is divided into N number of subgraph from left to right according to same dividing mode respectively, and N is the integer more than 1;
Threshold value acquisition module, each target pixel points in i-th of subgraph of the rainfall line image for finding out preliminary binaryzation
Corresponding pixel points in i-th of subgraph of low-resolution image, i=1,2 ..., N, and fall into a trap in CIE Lab color spaces
Calculate the maximum space distance dTh between these pixels and its average point (L_avg, a_avg, b_avg);
Rainfall line image output module, all pictures in i-th of subgraph of the digital picture for traversing rainfall record drawing
Vegetarian refreshments sets the pixel for meeting the following conditions to binaryzation target, and rest of pixels point is set as binaryzation background:In CIE
It is less than dTh with the distance between average point (L_avg, a_avg, b_avg) in Lab color spaces, and in HSI color spaces
H components are more than hTh, and hTh is the default H components threshold value more than or equal to 180;To obtain the rainfall line chart of final binaryzation
Picture.
6. the digitalizer of rainfall record drawing as claimed in claim 5, which is characterized in that the tilt corrector unit uses
Grid image that following methods export grid image extraction unit and rainfall line image extraction unit, rainfall line image carry out
Slant correction:The angle of inclination of longest straight line is detected using Hough transformation method from the rainfall line image of preliminary binaryzation,
Rotation correction finally is carried out to grid image, rainfall line image according to the angle of inclination detected.
7. the digitalizer of rainfall record drawing as claimed in claim 5, which is characterized in that the preliminary binary conversion treatment mould
Block carries out preliminary binary conversion treatment using improved Niblack algorithms to low-resolution image, specific as follows:It first will be described low
Image in different resolution gray processing;Then the binary-state threshold of each pixel is determined using following formula:
Wherein T (x, y) is the binary-state threshold of point (x, y), and m (x, y) is that the gray scale put in size template is preset around point (x, y)
Mean value;S (x, y) indicates the standard deviation of pixel gray value in template;The value range of parameter k is [0.4,0.6], R 128;
If the gray value at point (x, y) is less than T (x, y), then it is set and is otherwise set as target for background.
8. the digitalizer of rainfall record drawing as described in any one of claim 5~7, which is characterized in that the device also wraps
The image acquisition unit of the digital picture for obtaining rainfall record drawing is included, and defeated for storing rainfall computing unit institute
Go out the data storage cell of rainfall product data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510355671.4A CN106324708B (en) | 2015-06-24 | 2015-06-24 | Digitizing solution, the device of rainfall record drawing |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510355671.4A CN106324708B (en) | 2015-06-24 | 2015-06-24 | Digitizing solution, the device of rainfall record drawing |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106324708A CN106324708A (en) | 2017-01-11 |
CN106324708B true CN106324708B (en) | 2018-08-14 |
Family
ID=57729778
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510355671.4A Expired - Fee Related CN106324708B (en) | 2015-06-24 | 2015-06-24 | Digitizing solution, the device of rainfall record drawing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106324708B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107765348B (en) * | 2017-10-23 | 2020-06-02 | 成都信息工程大学 | Precipitation weather message data error correction method and rime landscape prediction method |
CN110533036B (en) * | 2019-08-28 | 2022-06-07 | 长城信息股份有限公司 | Rapid inclination correction method and system for bill scanned image |
CN111738208A (en) * | 2020-07-16 | 2020-10-02 | 江苏省气象信息中心 | Method for extracting tracks in meteorological self-recording paper |
CN113359216A (en) * | 2021-06-03 | 2021-09-07 | 山东捷瑞数字科技股份有限公司 | Method, system and storage medium for identification of recorded data of tipping-bucket rain gauge |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2010084521A1 (en) * | 2009-01-20 | 2010-07-29 | 本田技研工業株式会社 | Method and apparatus for identifying raindrops on a windshield |
US9258531B2 (en) * | 2009-07-27 | 2016-02-09 | Utc Fire & Security Corporation | System and method for video-quality enhancement |
CN101930594B (en) * | 2010-04-14 | 2012-05-23 | 山东山大鸥玛软件有限公司 | Rapid correction method for scanning document image |
-
2015
- 2015-06-24 CN CN201510355671.4A patent/CN106324708B/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN106324708A (en) | 2017-01-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111626190B (en) | Water level monitoring method for scale recognition based on clustering partition | |
Chen et al. | Isolating individual trees in a savanna woodland using small footprint lidar data | |
CN101599120B (en) | Identification method of remote sensing image building | |
CN108960198A (en) | A kind of road traffic sign detection and recognition methods based on residual error SSD model | |
CN103164692B (en) | A kind of special vehicle instrument automatic identification system based on computer vision and method | |
US9064151B2 (en) | Device and method for detecting plantation rows | |
CN107564017B (en) | Method for detecting and segmenting urban high-resolution remote sensing image shadow | |
CN107092871B (en) | Remote sensing image building detection method based on multiple dimensioned multiple features fusion | |
CN105205488B (en) | Word area detection method based on Harris angle points and stroke width | |
CN106324708B (en) | Digitizing solution, the device of rainfall record drawing | |
CN110569751B (en) | High-resolution remote sensing image building extraction method | |
CN109635733B (en) | Parking lot and vehicle target detection method based on visual saliency and queue correction | |
CN109376740A (en) | A kind of water gauge reading detection method based on video | |
CN110070545B (en) | Method for automatically extracting urban built-up area by urban texture feature density | |
CN107301649B (en) | Regional merged SAR image coastline detection algorithm based on superpixels | |
CN114596551A (en) | Vehicle-mounted forward-looking image crack detection method | |
CN114708208B (en) | Machine vision-based famous tea tender bud identification and picking point positioning method | |
CN113609984A (en) | Pointer instrument reading identification method and device and electronic equipment | |
CN113409332B (en) | Building plane segmentation method based on three-dimensional point cloud | |
CN112085725B (en) | Residual film residual quantity detection method and early warning system based on heuristic iterative algorithm | |
CN111882573B (en) | Cultivated land block extraction method and system based on high-resolution image data | |
CN109815784A (en) | A kind of intelligent method for classifying based on thermal infrared imager, system and storage medium | |
CN103065296B (en) | High-resolution remote sensing image residential area extraction method based on edge feature | |
CN107657246A (en) | A kind of remote sensing image building detection method based on multi-scale filtering building index | |
CN103366376B (en) | A kind of image characteristic extracting method based on neighborhood dimensional variation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20180629 Address after: 210036 room 3-1302, moonlight Plaza, Longjiang District, Gulou District, Nanjing, Jiangsu. Applicant after: Hao Li Applicant after: Li Shijin Address before: 210036 room 3-1302, moonlight Plaza, Longjiang District, Gulou District, Nanjing, Jiangsu. Applicant before: Li Shijin |
|
TA01 | Transfer of patent application right | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180814 Termination date: 20200624 |
|
CF01 | Termination of patent right due to non-payment of annual fee |