CN107492106A - A kind of water-surface areas dividing method - Google Patents
A kind of water-surface areas dividing method Download PDFInfo
- Publication number
- CN107492106A CN107492106A CN201710740818.0A CN201710740818A CN107492106A CN 107492106 A CN107492106 A CN 107492106A CN 201710740818 A CN201710740818 A CN 201710740818A CN 107492106 A CN107492106 A CN 107492106A
- Authority
- CN
- China
- Prior art keywords
- image
- grid
- gray level
- water
- sample
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/155—Segmentation; Edge detection involving morphological operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
- G06T7/41—Analysis of texture based on statistical description of texture
- G06T7/45—Analysis of texture based on statistical description of texture using co-occurrence matrix computation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Probability & Statistics with Applications (AREA)
- Computing Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of water-surface areas dividing method, including：Grid image will be generated after gray level image gridding to be split；Image texture distribution map is obtained after carrying out rim detection, adaptive Threshold and closing operation of mathematical morphology to the gray level image；Sample areas using the central area of each grid in the grid image as sparse sampling, the textural characteristics in each sample grid region are calculated using gray level co-occurrence matrixes, draw feature vector chart；The feature vector chart is judged using supporting vector machine model and generates judgement grid image；Projection histogram and horizontal meaders technology, generation sample aqua region section are used to the judgement grid image；The filling that floods is carried out on the texture density profile according to the sample aqua region section, obtains pool area image.Using embodiments of the invention can remove in shooting process to picture strip come distortion effects, speed-up computation speed and more accurate Ground Split water-surface areas.
Description
Technical field
The present invention relates to image procossing and image segmentation field, more particularly to a kind of method of water-surface areas segmentation.
Background technology
With the continuous development of computer science and technology, image procossing and image segmentation have gradually formed the science body of oneself
System, application field also constantly expand.The existing similar techniques of generally use are mostly in automatic driving car in existing water surface cutting techniques
Carried out in the tilt system of camera, the method based on brightness and saturation degree feature, using sky as reference, in saturation
Spend in brightness space using sky as reference, found by sky in the inverted image of the water surface or directly using the saturation degree of sky as label
The water surface, while detected using RGB with HSV space in COLOR COMPOSITION THROUGH DISTRIBUTION, brightness and saturation degree are taken in HSV color spaces
Color characteristic of the ratio as water body, differentiate water using SVMs and sentence with non-aqueous region aids, obtain the figure of water-surface areas
As judging.But the determination methods poor reliability of brightness, saturation degree and color is based only in the prior art, and by illumination and water
Body turbidity influences greatly, to cause distortion factor and noise more, is not suitable for the situation for having the shadow of the trees, illumination mutability and water pollution.Only
The accuracy of judgement water-surface areas is cannot ensure as the method judged using only SVMs.
The content of the invention
The present invention provides a kind of water-surface areas dividing method, can remove the distortion shadow come in shooting process to picture strip
Sound, speed-up computation speed and more accurate Ground Split water-surface areas.
One aspect of the present invention provides a kind of water-surface areas dividing method, including：
S11, grid image will be generated after the gray level image gridding in region to be split；Using Canny operators to the ash
Spend image and carry out rim detection, generate texture density profile, it is true to perform adaptive threshold value to the texture density profile
Image texture distribution map is obtained after determining method and closing operation of mathematical morphology；The size of each grid is N*N, N in the grid image
For integer, and N is more than or equal to 1；
S12, the sample areas using the central area of each grid in the grid image as sparse sampling, use gray scale
Co-occurrence matrix calculates the textural characteristics of each sample areas, draws feature vector chart corresponding to each sample areas；
S13, each feature vector chart is judged using supporting vector machine model and generates judgement grid image；
S14, projection histogram and horizontal meaders technology are used to each judgement grid image, generate sample aqua region
Section；
S15, the filling that floods is carried out on described image grain distribution figure according to the sample aqua region section, obtain pool
Area image.
A kind of water-surface areas dividing method provided in an embodiment of the present invention to gray level image to be split by carrying out edge
Image texture distribution map is obtained after detection, adaptive Threshold and closing operation of mathematical morphology；Again by by the gray scale
Calculated after image gridding and sparse sampling with gray level co-occurrence matrixes, obtain feature vector chart, solve in the prior art
Brightness, saturation degree and color are based only on to carry out the problem of judgement causes reliability relatively low, the feature vector chart is to judging
Whether water-surface areas has a higher reference value, and carries out calculating using the gray level co-occurrence matrixes and can accelerate to calculate speed
Degree；Meanwhile by result of determination and described image grain distribution figure after being judged with SVMs the feature vector chart
The filling that floods is carried out, obtains pool area image, can quickly carry out judging whether water-surface areas, and more accurately split water
Face region.
As the improvement of this programme, also include before the step S11：
S101, original color image is changed into gray level image；
S102, two-dimensional coordinate system is established to the gray level image, and take the image in Y-direction in preset range as ash
Image is spent, the preset range is the distortion part of the gray level image.
A kind of water-surface areas dividing method provided in an embodiment of the present invention is built by the way that original image is switched into gray level image
Vertical two-dimensional coordinate system removes the distortion part of image, can remove the distortion effects come in shooting process to picture strip, remain with
Gray level image.
As the improvement of this programme, in the step S11, the edge pixel of the rim detection is 255, utilizes convolution kernel
Gray level image after the rim detection is summed, generates texture density profile.
As the improvement of this programme, in the step S12, the feature vector chart includes 16 dimensional feature vectors；The ash
Degree co-occurrence matrix four characteristic values draw 16 dimensional feature vector in the feature of four angles respectively, generate the feature to
Spirogram；Four angles include 0 °, 45 °, 90 ° and 135 °；Four characteristic values include energy, entropy, contrast and homogeney.
As the improvement of this programme, in the step S12, the sparse sampling is by selecting each of the grid image
The central area of grid is as sample areas, and the sample areas size is M*M, and M is integer, and M is more than or equal to 1, and M is small
In N.
As the improvement of this programme, the step S14 includes：
S141, sampled point discrete in the feature vector chart is merged by the way of projection histogram, and generate water
Region interval graph；
S142, horizontal meaders removal discontinuity zone and independent noise spot, generation are carried out to the aqua region interval graph
Sample aqua region section.
A kind of water-surface areas dividing method provided in an embodiment of the present invention passes through projection histogram and horizontal meaders technology pair
The feature vector chart removes discontinuity zone and noise, solves and image is caused by illumination and water body muddiness in the prior art
Influence, apparent sample aqua region section can be obtained.
As the improvement of this programme, in the step S15, the filling effect that floods in described image grain distribution figure,
Each grid element center point in the aqua region section starts simultaneously to enter in described image grain distribution figure as seed point
The capable filling that floods, the size of the central point determine according to the image size in region to be split；During the filling that floods is all when starting
Heart point starts simultaneously at filling.
As the improvement of this programme, the supporting vector machine model used in the step S13 is by training in advance
Obtain, the training method of the supporting vector machine model includes：
S21, grid image will be generated after the gray level image gridding in region to be split；The grid image is carried out artificial
Water-surface areas is determine whether, generates artificial determinating area figure；The size of a grid is N*N in the grid image, and N is whole
Number, and N is more than or equal to 1；
S22, the sample areas using each grid in the grid image as intensive sampling, use gray level co-occurrence matrixes meter
The textural characteristics in each sample grid region are calculated, draw feature vector chart corresponding to each sample areas；
S23, the feature vector chart is combined with the artificial determinating area figure, draws the SVMs mould
Type.
A kind of water-surface areas dividing method provided in an embodiment of the present invention draws feature vector chart by gray level co-occurrence matrixes,
And the technology being combined with artificial determinating area figure, supporting vector machine model is obtained, can more accurately judge that water-surface areas.
As the improvement of this programme, also include before methods described S21：
S201, original color image is changed into gray level image；
S202, two-dimensional coordinate system is established to the gray level image, and take the image in Y-direction in preset range as ash
Image is spent, the preset range is the distortion part of the gray level image.
As the improvement of this programme, feature vector chart described in methods described S22 includes 16 dimensional feature vectors；The gray scale
Four characteristic values of co-occurrence matrix draw 16 dimensional feature vector in the feature of four angles respectively, generate the characteristic vector
Figure；Four angles include 0 °, 45 °, 90 ° and 135 °；Four characteristic values include energy, entropy, contrast and homogeney.
Brief description of the drawings
Fig. 1 is a kind of schematic flow sheet of water-surface areas dividing method in the embodiment of the present invention；
Fig. 2 is the schematic diagram of grid image in the embodiment of the present invention；
Fig. 3 be flooded in the embodiment of the present invention filling central point schematic diagram；
Fig. 4 is a kind of schematic flow sheet of water-surface areas dividing method in the embodiment of the present invention；
Fig. 5 is a kind of schematic flow sheet of water-surface areas dividing method in the embodiment of the present invention；
Fig. 6 is the schematic flow sheet of the training method of supporting vector machine model in the embodiment of the present invention；
Fig. 7 is the schematic flow sheet of the training method of supporting vector machine model in the embodiment of the present invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other under the premise of creative work is not made
Embodiment, belong to the scope of protection of the invention.
Referring to Fig. 1, a kind of Fig. 1 schematic flow sheets of water-surface areas dividing method provided in an embodiment of the present invention, including：
S11, grid image will be generated after the gray level image gridding in region to be split；Using Canny operators to the ash
Spend image and carry out rim detection, generate texture density profile, it is true to perform adaptive threshold value to the texture density profile
Image texture distribution map is obtained after determining method and closing operation of mathematical morphology；The size of each grid is N*N, N in the grid image
For integer, and N is more than or equal to 1；
S12, the sample areas using the central area of each grid in the grid image as sparse sampling, use gray scale
Co-occurrence matrix calculates the textural characteristics of each sample areas, draws feature vector chart corresponding to each sample areas；
S13, each feature vector chart is judged using supporting vector machine model and generates judgement grid image；
S14, projection histogram and horizontal meaders technology are used to each judgement grid image, generate sample aqua region
Section；
S15, the filling that floods is carried out on described image grain distribution figure according to the sample aqua region section, obtain pool
Area image.
Wherein, in the step S11, the edge pixel of the rim detection is 255, and the edge is examined using convolution kernel
Gray level image summation after survey, generates texture density profile.
Wherein, in the step S12, the feature vector chart includes 16 dimensional feature vectors；The gray level co-occurrence matrixes
Four characteristic values draw 16 dimensional feature vector in the feature of four angles respectively, generate the feature vector chart；Described four
Individual angle includes 0 °, 45 °, 90 ° and 135 °；Four characteristic values include energy, entropy, contrast and homogeney.
Wherein, sparse sampling described in the step S12 is by selecting the central area of each grid of the grid image
As sample areas, the sample areas size is M*M, and M is integer, and M is more than or equal to 1, and M is less than N.
Specifically, grid image is generated with N*N pixel size grid divisions to gray level image to be split；Such as Fig. 2 institutes
Show, the size of each grid is N*N in the grid image, and N is integer, and N is more than or equal to 1；Specifically, to gray level image
Rim detection is carried out using Canny operators, it is 255 to set the edge pixel detected, and with convolution kernel to the gray level image
Summed, obtain texture density profile, the value of the convolution sum is all 1；Specifically, the texture density profile is held
The adaptive Threshold of row, the gray level image is chosen by appropriate threshold value still can reflect figure so as to obtain
As overall and local feature binary image, image texture distribution map, the form are obtained after recycling closing operation of mathematical morphology
Learn closed operation and the border of each 1 pixel coordinator of the binary image is expanded one layer, then each 1 picture of the binary image
The boundary point of plain coordinator removes so as to reduce one layer, and the hole in target is removed using the method for first expanding post-etching, so as to
Obtain image texture distribution map.
Specifically, using the central area of each grid in the grid image as the sample areas of sparse sampling, such as Fig. 2
Shown, the black bars in figure are the sample areas, and the sample areas size is M*M, and M is integer, and M is more than or waited
In 1, and M is less than N；Specifically, calculate gray scale symbiosis square of the sample areas in 0 °, 45 °, 90 ° and 135 ° four angles
Battle array, and generates four energy, entropy, contrast and homogeney features, by four, each direction of four direction feature, 16 features altogether
16 dimensional feature vectors are formed, generate 16 dimensional feature vector figures.
Specifically, judging using supporting vector machine model each feature vector chart, and generate judgement grid
Image；
Specifically, using projection histogram technology and horizontal meaders technology to each judgement grid image, sample is generated
This aqua region section；
Specifically, carrying out the filling that floods on described image grain distribution figure according to the sample aqua region section, obtain
Pool area image；Specifically, the filling effect that floods is in described image grain distribution figure, according in the aqua region section
Each grid element center point starts simultaneously at as seed point carries out the filling that floods in described image grain distribution figure, the central point
Size determines that all central points start simultaneously at filling when the filling that floods starts, in described according to the image size in region to be split
The schematic diagram of heart point is as shown in figure 3, the white square in black bars is the central point.
A kind of water-surface areas dividing method provided in an embodiment of the present invention to gray level image to be split by carrying out edge
Image texture distribution map is obtained after detection, adaptive Threshold and closing operation of mathematical morphology；Again by by the gray scale
Calculated after image gridding and sparse sampling with gray level co-occurrence matrixes, obtain feature vector chart, solve in the prior art
Brightness, saturation degree and color are based only on to carry out the problem of judgement causes reliability relatively low, the feature vector chart is to judging
Whether water-surface areas has a higher reference value, and carries out calculating using the gray level co-occurrence matrixes and can accelerate to calculate speed
Degree；Meanwhile by result of determination and described image grain distribution figure after being judged with SVMs the feature vector chart
The filling that floods is carried out, obtains aqua region figure, can quickly carry out judging whether water-surface areas, and more accurately split the water surface
Region.
Referring to Fig. 4, Fig. 4 is a kind of schematic flow sheet of water-surface areas dividing method in the embodiment of the present invention, including：
S101, original color image is changed into gray level image；
S102, two-dimensional coordinate system is established to the gray level image, and take the image in Y-direction in preset range as ash
Image is spent, the preset range is the distortion part of the gray level image.
Specifically, reading raw video image by frame, camera shake and camera are removed in itself using median filtering method
The noise that imperfect tape comes, gray level image is converted to by original RGB color image.
Specifically, it is origin to define the image upper left corner, right direction is X-axis, and lower direction is Y-axis, and X-axis is row, and Y-axis is row,
Establish two-dimensional coordinate system；Shoot towards ground because unmanned plane image is generally vertical or tilted with the angle of depression more than -45 ° towards ground
Face is shot, and aqua region generally longitudinally runs through image when shooting, during in order to remove tilt image distortion and removal contain
The top half image of a large amount of meaningless elements, takes the pending ribbon image-region after preset range is in Y-direction to make
For the gray level image.
A kind of water-surface areas dividing method provided in an embodiment of the present invention is built by the way that original image is switched into gray level image
Vertical two-dimensional coordinate system removes the distortion part of image, can remove the distortion effects come in shooting process to picture strip, remain with
Gray level image.
Referring to Fig. 5, Fig. 5 is a kind of schematic flow sheet of water-surface areas dividing method in the embodiment of the present invention, including：
S141, sampled point discrete in the feature vector chart is merged by the way of projection histogram, and generate water
Region interval graph；
S142, horizontal meaders removal discontinuity zone and independent noise spot, generation are carried out to the aqua region interval graph
Sample aqua region section.
Specifically, merged sampled point discrete in the feature vector chart by the way of projection histogram, will be each
The result of determination of sample is arranged as two-dimensional matrix according to sample region position, and according to each row of longitudinal direction statistics summation
The sample point quantity of water is judged as, generation have recorded the one-dimensional statistics array of each row water sample quantity, generation aqua region area
Between scheme.
Specifically, carrying out horizontal meaders to the aqua region interval graph removes discontinuity zone and independent noise spot, institute
The region of region and value more than 0 that value is 0 can be contained by stating one-dimensional statistics array, and the region being worth for 0 represents not to be had on this row
Any aqua region, it is to represent to include possible aqua region on this row that value, which is more than 0,；Part non-zero value may in one-dimensional statistics array
Can be two parts by a single 0 value cutting because of a shadow band in original image, it is also possible to because original
There is the situation of an isolated non-zero value in a fritter noise in image, therefore, now removes what is isolated in one-dimensional statistics array
0 value and isolated non-zero value, calculate continuous non-zero value piecewise interval；Statistical counting maximum in section is less than possible maximum
2/3rds section of value is directly given up, and the section of this type is not generally longitudinally through whole ribbon image-region, therefore
Need to remove, remaining section is the sample aqua region section.
Referring to Fig. 6, Fig. 6 is the schematic flow sheet of the training method of supporting vector machine model in the embodiment of the present invention, including：
S21, grid image will be generated after the gray level image gridding in region to be split；The grid image is carried out artificial
Water-surface areas is determine whether, generates artificial determinating area figure；The size of a grid is N*N in the grid image, and N is whole
Number, and N is more than or equal to 1；
S22, the sample areas using each grid in the grid image as intensive sampling, use gray level co-occurrence matrixes meter
The textural characteristics in each sample grid region are calculated, draw feature vector chart corresponding to each sample areas；
S23, the feature vector chart is combined with the artificial determinating area figure, draws the SVMs mould
Type.
Wherein, in the step S22, the feature vector chart includes 16 dimensional feature vectors；The gray level co-occurrence matrixes
Four characteristic values draw 16 dimensional feature vector in the feature of four angles respectively, generate the feature vector chart；Described four
Individual angle includes 0 °, 45 °, 90 ° and 135 °；Four characteristic values include energy, entropy, contrast and homogeney.
Specifically, grid image is generated with N*N pixel size grid divisions to gray level image to be split；To the net
Table images carry out manually determining whether water-surface areas, generate artificial determinating area figure；Each grid in the grid image
Size is N*N, and N is integer, and N is more than or equal to 1.
Specifically, using each grid in the grid image as the sample areas of intensive sampling, the sample area is calculated
Gray level co-occurrence matrixes of the domain in 0 °, 45 °, 90 ° and 135 ° four angles, and generate four energy, entropy, contrast and homogeney spies
Sign, by four, each direction of four direction feature, 16 features form 16 dimensional feature vectors altogether, generate 16 dimensional feature vector figures.
Specifically, the feature vector chart is combined with the artificial determinating area figure, the SVMs is drawn
Model, the supporting vector machine model include 16 dimensional feature vector.
A kind of water-surface areas dividing method provided in an embodiment of the present invention draws feature vector chart by gray level co-occurrence matrixes,
And the technology being combined with artificial determinating area figure, the supporting vector machine model is obtained, can more accurately judge that the water surface
Region.
Referring to Fig. 7, Fig. 7 is the schematic flow sheet of the training method of supporting vector machine model in the embodiment of the present invention, including：
S201, original color image is changed into gray level image；
S202, two-dimensional coordinate system is established to the gray level image, and take the image in Y-direction in preset range as ash
Image is spent, the preset range is the distortion part of the gray level image.
Specifically, reading raw video image by frame, camera shake and camera are removed in itself using median filtering method
The noise that imperfect tape comes, gray level image is converted to by original RGB color image.
Specifically, it is origin to define the image upper left corner, right direction is X-axis, and lower direction is Y-axis, and X-axis is row, and Y-axis is row,
Establish two-dimensional coordinate system；Shoot towards ground because unmanned plane image is generally vertical or tilted with the angle of depression more than -45 ° towards ground
Face is shot, and aqua region generally longitudinally runs through image when shooting, during in order to remove tilt image distortion and removal contain
The top half image of a large amount of meaningless elements, takes the pending ribbon image-region after preset range is in Y-direction to make
For the gray level image.
A kind of water-surface areas dividing method provided in an embodiment of the present invention is built by the way that original image is switched into gray level image
Vertical two-dimensional coordinate system removes the distortion part of image, carrys out Training Support Vector Machines with this and obtains the supporting vector machine model, energy
The distortion effects come in shooting process to picture strip are enough removed, the gray level image remained with, make the supporting vector model
Judge better.
In summary, a kind of water-surface areas dividing method provided in an embodiment of the present invention passes through to gray level image to be split
Image texture distribution map is obtained after carrying out rim detection, adaptive Threshold and closing operation of mathematical morphology；Pass through again by
Calculated after the gray level image gridding and sparse sampling with gray level co-occurrence matrixes, obtain feature vector chart, solved existing
Have and brightness, saturation degree and color are based only in technology to carry out the problem of judgement causes reliability relatively low, the characteristic vector
Figure can add to judging whether that water-surface areas has higher reference value, and carrying out calculating using the gray level co-occurrence matrixes
Fast calculating speed；Meanwhile by result of determination and described image line after being judged with SVMs the feature vector chart
Reason distribution map carries out the filling that floods, and obtains pool area image, can quickly carry out judging whether water-surface areas, and more accurate
Segmentation water-surface areas.
Described above is the preferred embodiment of the present invention, it is noted that for those skilled in the art
For, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improvements and modifications are also considered as
Protection scope of the present invention.
Claims (10)
1. a kind of water-surface areas dividing method, it is characterised in that methods described includes：
S11, grid image will be generated after the gray level image gridding in region to be split；Using Canny operators to the gray-scale map
As carrying out rim detection, texture density profile is generated, adaptive threshold value determination side is performed to the texture density profile
Image texture distribution map is obtained after method and closing operation of mathematical morphology；The size of each grid is N*N in the grid image, and N is whole
Number, and N is more than or equal to 1；
S12, the sample areas using the central area of each grid in the grid image as sparse sampling, use gray scale symbiosis
The textural characteristics of each sample areas of matrix computations, draw feature vector chart corresponding to each sample areas；
S13, each feature vector chart is judged using supporting vector machine model and generates judgement grid image；
S14, projection histogram and horizontal meaders technology, generation sample aqua region area are used to each judgement grid image
Between；
S15, the filling that floods is carried out on described image grain distribution figure according to the sample aqua region section, obtain aqua region figure
Picture.
2. water-surface areas dividing method according to claim 1, it is characterised in that also include before the step S11：
S101, original color image is changed into gray level image；
S102, two-dimensional coordinate system is established to the gray level image, and take the image in Y-direction in preset range as gray-scale map
Picture, the preset range are the distortion part of the gray level image.
3. water-surface areas dividing method according to claim 1, it is characterised in that in the step S11, the edge inspection
The edge pixel of survey is 255, and the gray level image after the rim detection is summed using convolution kernel, generates the texture density point
Butut.
4. water-surface areas dividing method according to claim 1, it is characterised in that in the step S12, the feature to
Spirogram includes 16 dimensional feature vectors；Four characteristic values of the gray level co-occurrence matrixes draw described in the feature of four angles respectively
16 dimensional feature vectors, generate the feature vector chart；Four angles include 0 °, 45 °, 90 ° and 135 °；Four features
Value includes energy, entropy, contrast and homogeney.
5. water-surface areas dividing method according to claim 1, it is characterised in that described sparse to adopt in the step S12
For sample by selecting the central area of each grid of the grid image to be used as sample areas, the sample areas size is M*M,
M is integer, and M is more than or equal to 1, and M is less than N.
6. water-surface areas dividing method according to claim 1, it is characterised in that the step S14 includes：
S141, sampled point discrete in the feature vector chart is merged by the way of projection histogram, and generate aqua region
Interval graph；
S142, horizontal meaders removal discontinuity zone and independent noise spot are carried out to the aqua region interval graph, generate sample
Aqua region section.
7. water-surface areas dividing method according to claim 1, it is characterised in that in the step S15, described flood is filled out
Use as and be used for described image grain distribution figure, each grid element center point in the aqua region section as seed point simultaneously
Start to carry out the filling that floods in described image grain distribution figure, the size of the central point is according to the image size in region to be split
Determine；All central points start simultaneously at filling when the filling that floods starts.
8. water-surface areas dividing method according to claim 1, it is characterised in that used in the step S13 described
Supporting vector machine model obtains by training in advance, and the training method of the supporting vector machine model includes：
S21, grid image will be generated after the gray level image gridding in region to be split；The grid image is manually judged
Whether it is water-surface areas, generates artificial determinating area figure；The size of a grid is N*N in the grid image, and N is integer,
And N is more than or equal to 1；
S22, the sample areas using each grid in the grid image as intensive sampling, calculated using gray level co-occurrence matrixes every
The textural characteristics in the individual sample grid region, draw feature vector chart corresponding to each sample areas；
S23, the feature vector chart is combined with the artificial determinating area figure, draws the supporting vector machine model.
9. water-surface areas dividing method according to claim 8, it is characterised in that also include before the step S21：
S201, original color image is changed into gray level image；
S202, two-dimensional coordinate system is established to the gray level image, and take the image in Y-direction in preset range as gray-scale map
Picture, the preset range are the distortion part of the gray level image.
10. water-surface areas dividing method according to claim 8, it is characterised in that in the step S22, the feature
Vectogram includes 16 dimensional feature vectors；Four characteristic values of the gray level co-occurrence matrixes draw institute in the feature of four angles respectively
16 dimensional feature vectors are stated, generate the feature vector chart；Four angles include 0 °, 45 °, 90 ° and 135 °；Four spies
Value indicative includes energy, entropy, contrast and homogeney.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710740818.0A CN107492106A (en) | 2017-08-24 | 2017-08-24 | A kind of water-surface areas dividing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710740818.0A CN107492106A (en) | 2017-08-24 | 2017-08-24 | A kind of water-surface areas dividing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107492106A true CN107492106A (en) | 2017-12-19 |
Family
ID=60645710
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710740818.0A Pending CN107492106A (en) | 2017-08-24 | 2017-08-24 | A kind of water-surface areas dividing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107492106A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109255792A (en) * | 2018-08-02 | 2019-01-22 | 广州市鑫广飞信息科技有限公司 | A kind of dividing method of video image, device, terminal device and storage medium |
| CN110428386A (en) * | 2019-06-25 | 2019-11-08 | 口口相传(北京)网络技术有限公司 | Map grid merging method, device, storage medium, electronic device |
| CN110458019A (en) * | 2019-07-10 | 2019-11-15 | 哈尔滨工程大学 | The waterborne target detection method for excluding inverted image interference under rare cognition sample conditions |
| CN111220786A (en) * | 2020-03-09 | 2020-06-02 | 生态环境部华南环境科学研究所 | Method for rapidly monitoring organic pollution of deep water sediments |
| CN112893186A (en) * | 2021-01-13 | 2021-06-04 | 山西能源学院 | Method and system for quickly visually detecting power-on of LED lamp filament |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102054268A (en) * | 2011-01-06 | 2011-05-11 | 西安电子科技大学 | Adaptive segmentation method of SAR (Stop and Reveres) image water area |
| CN103325110A (en) * | 2013-05-29 | 2013-09-25 | 山西绿色光电产业科学技术研究院（有限公司） | Panoramic image correction algorithm based on panoramic all-in-one speed dome camera |
-
2017
- 2017-08-24 CN CN201710740818.0A patent/CN107492106A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102054268A (en) * | 2011-01-06 | 2011-05-11 | 西安电子科技大学 | Adaptive segmentation method of SAR (Stop and Reveres) image water area |
| CN103325110A (en) * | 2013-05-29 | 2013-09-25 | 山西绿色光电产业科学技术研究院（有限公司） | Panoramic image correction algorithm based on panoramic all-in-one speed dome camera |
Non-Patent Citations (4)
| Title |
|---|
| 刘凯: "电视图像中的字幕检测算法", 《中国优秀硕士学位论文全文数据库 信息科技辑 ISSN 1674-0246》 * |
| 梁青青: "超市货架区域分割与商品识别研究", 《中国优秀硕士学位论文全文数据库 信息科技辑 I138-1174》 * |
| 沈叶健等: "一种有效的可见光图像中水坝目标的识别方法", 《计算机应用》 * |
| 胡蕾等: "一种基于光谱与纹理特征的多光谱遥感图像地物分类方法", 《中国科技论文》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109255792A (en) * | 2018-08-02 | 2019-01-22 | 广州市鑫广飞信息科技有限公司 | A kind of dividing method of video image, device, terminal device and storage medium |
| CN110428386A (en) * | 2019-06-25 | 2019-11-08 | 口口相传(北京)网络技术有限公司 | Map grid merging method, device, storage medium, electronic device |
| CN110458019A (en) * | 2019-07-10 | 2019-11-15 | 哈尔滨工程大学 | The waterborne target detection method for excluding inverted image interference under rare cognition sample conditions |
| CN111220786A (en) * | 2020-03-09 | 2020-06-02 | 生态环境部华南环境科学研究所 | Method for rapidly monitoring organic pollution of deep water sediments |
| CN112893186A (en) * | 2021-01-13 | 2021-06-04 | 山西能源学院 | Method and system for quickly visually detecting power-on of LED lamp filament |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107492106A (en) | A kind of water-surface areas dividing method | |
| CN109978839B (en) | Method for detecting wafer low-texture defects | |
| CN102682446B (en) | Adaptive combined two-sided filter is used to generate equipment and the method for dense depth map | |
| EP0394929B1 (en) | Dither image processing method | |
| KR101617059B1 (en) | Method and system for image resizing based on interpolation enhanced seam operations | |
| US8374428B2 (en) | Color balancing for partially overlapping images | |
| US9922443B2 (en) | Texturing a three-dimensional scanned model with localized patch colors | |
| CN110544300B (en) | Method for automatically generating three-dimensional model based on two-dimensional hand-drawn image characteristics | |
| CN105096330A (en) | Image processing method capable of automatically recognizing pure-color borders, system and a photographing terminal | |
| CN115170669A (en) | Identification and positioning method and system based on edge feature point set registration and storage medium | |
| JP4428936B2 (en) | Method for obtaining Euclidean distance from point in 3D space to 3D object surface from projection distance image stored in memory having projection distance along projection direction from projection plane to 3D object surface as pixel value | |
| CN104504703A (en) | Welding spot color image segmentation method based on chip element SMT (surface mounting technology) | |
| CN106780504B (en) | It is a kind of based on distance keep level set beef image in carry on the back long flesh automatic division method | |
| CN109064439B (en) | Partition-based single-side light-entering type light guide plate shadow defect extraction method | |
| CN111753693B (en) | Target detection method under static scene | |
| CN111489322A (en) | Method and device for adding sky filter to static picture | |
| CN108629742B (en) | True ortho image shadow detection and compensation method, device and storage medium | |
| CN116222381A (en) | Electrode coating size measurement method and device | |
| CN113658144B (en) | Method, device, equipment and medium for determining geometric information of pavement diseases | |
| CN106469432A (en) | Object extraction method and object extraction equipment | |
| CN108109120B (en) | Illumination compensation method and device for dot matrix two-dimensional code | |
| CN107702885B (en) | Bed Particle plays the extracting method of outstanding time | |
| CN115601616A (en) | Sample data generation method and device, electronic equipment and storage medium | |
| CN110619338B (en) | Image feature extraction method capable of long-time dependence | |
| Chen et al. | Candidate region acquisition optimization algorithm based on multi-granularity data enhancement |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171219 |
|
| RJ01 | Rejection of invention patent application after publication |