CN109242870A - A kind of sea horizon detection method divided based on image with textural characteristics - Google Patents
A kind of sea horizon detection method divided based on image with textural characteristics Download PDFInfo
- Publication number
- CN109242870A CN109242870A CN201810768241.9A CN201810768241A CN109242870A CN 109242870 A CN109242870 A CN 109242870A CN 201810768241 A CN201810768241 A CN 201810768241A CN 109242870 A CN109242870 A CN 109242870A
- Authority
- CN
- China
- Prior art keywords
- image
- region
- value
- pixel
- sea horizon
- 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
- G06T7/13—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
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
Abstract
The invention discloses a kind of sea horizon detection methods divided based on image with textural characteristics, comprising the following steps: 1) obtains colored sea image；2) original color image is converted into gray level image；3) gaussian filtering process is carried out to gray level image using gaussian filtering frame；4) image filtered is evenly dividing along the vertical direction calculate separately out for several regions each region direction vertically downward, the gray level co-occurrence matrixes that distance is 1 between point pair and its contrast level parameter；5) the ratio between the co-occurrence matrix contrast in every piece of region and upper one piece of region is calculated, wherein the maximum region of ratio is sea horizon region；6) gradient for calculating selected areas vertical direction, finds out threshold value using maximum variance between clusters, and the pixel that gradient value is greater than threshold value is carried out straight-line detection as candidate point；7) Hough transform straight-line detection is carried out, the best straight line detected is sea horizon.Sea horizon position in the image of sea can be effectively detected out in method of the invention.
Description
Technical field
The present invention relates to technical field of image processing, and in particular to a kind of sea horizon divided based on image with textural characteristics
Detection method.
Background technique
Sea horizon detection is the important link of sail and sea unmanned probing.As unmanned boat patrols He Hai in sea boat
Upper search and rescue etc. are more and more widely used, and image processing techniques is also applied to sea detection more and more.Wherein, sea horizon
Detection is of great significance in the detection of sea.Sea image can be divided into seawater and sky two by sea horizon detection
Part reduces search range for further progress target detection, this is for reducing the calculation amount of follow-up work with very heavy
The meaning wanted.Meanwhile image-region is divided into two parts by sea horizon detection, can eliminate sky cloud layer, haze and sea in image
The influence of background etc. on the bank improves the accuracy of target detection.
Relatively common sea horizon detection method is at present: by being handled to obtain an institute in it to entire image
There is element to meet the method that the point set recycled to the point set of provisioning request carries out best straight line fitting and obtains sea horizon.Often
The method seen is that the edge feature of image is extracted by canny algorithm (or sobel algorithm), and the threshold value in algorithm is as judgement figure
Whether the pixel of picture meets point set specified criteria partition value, and edge feature, which is significantly put, will be classified as candidate point.It is logical
Cross maximum between-cluster variance (Otsu) algorithm threshold value size.It is finally fitted to obtain sea horizon, such as Hough using obtained point set
Converter technique, least square method and random sampling coherence method (RANSAC) etc..But these methods are all by whole picture figure
As carry out global search judge to obtain candidate point, the selection of candidate point and the edge feature of image are related, thus sea ripple with
Many noise spots are chosen as candidate point in sky cloud layer, produce the sea horizon being finally fitted in obtained sea horizon and real image
Raw deviation is even fitted obtained sea horizon and is distributed between cloud layer or ripple.
Summary of the invention
In view of the deficiencies of the prior art, the present invention provides a kind of sea horizon detection sides divided based on image with textural characteristics
Sea horizon position can be effectively detected in method.
In order to achieve the above objectives, the present invention adopts the following technical scheme:
A kind of sea horizon detection method divided based on image with textural characteristics, comprising the following steps:
1) original colored sea image is obtained；
2) original color image is converted into gray level image；
3) gaussian filtering process is carried out to gray level image using gaussian filtering frame；
4) image filtered is evenly dividing along the vertical direction as several regions, the tonal gradation in each region is dropped
It is 16 grades, calculates separately out each region co-occurrence matrix that distance is 1 between vertically downward direction, point pair, and calculates each total
The contrast level parameter of raw matrix；
5) it is supreme down, calculate each piece of region and upper one piece of region the ratio between co-occurrence matrix contrast, wherein being compared
It is worth maximum region as the region where sea horizon；
6) gradient magnitude of region two pieces of region vertical direction adjacent thereto where sea horizon is calculated, with each pixel
The gradient magnitude of point vertical direction calculates the threshold size of this group of data using maximum variance between clusters as initial data,
Pixel using vertical gradient value greater than threshold value is cast out less than the pixel of threshold value and is not considered as candidate point；
7) Hough transform straight-line detection is carried out using obtained candidate point, the best straight line detected is sea horizon.
The present invention has the beneficial effect that compared with art methods
1, the method that the present invention is divided using image, divides an image into several regions, co-occurrence matrix is to describe every piece
The textural characteristics in region substantially determine the position where sea horizon by comparing co-occurrence matrix, only to sea horizon rough estimate institute
Subsequent processing is carried out in region, greatly reduces the influence of sky cloud layer and sea ripple, while only estimating roughly to sea horizon
Meter region is handled the calculation amount for also greatly reducing follow-up work.
2, the present invention avoids the dry of horizontal direction texture using the method for calculating selected areas vertical gradient value
It disturbs；It selects maximum variance between clusters to select candidate point simultaneously, keeps the probability of a classification mistake minute minimum.
Detailed description of the invention
Fig. 1 is the flow chart of the method for the present invention.
Fig. 2 is the schematic diagram of the method for the present invention one embodiment, wherein (a) is original image；It (b) is gray level image；(c)
To pass through filtered image；It (d) is the image after region division；It (e) is the maximum image of contrast region of variation；
It (f) is sea horizon position rough estimate figure；(g) sea horizon testing result figure.
Specific embodiment
It is clear to be more clear the object, technical solutions and advantages of the present invention, with reference to the accompanying drawing, to tool of the invention
Body embodiment elaborates.
As shown in Figure 1, a kind of sea horizon detection method divided based on image with textural characteristics, comprising the following steps:
1) original colored sea image is obtained；
2) original color image is converted into gray level image；
3) gaussian filtering process is carried out to gray level image using gaussian filtering frame；
4) image filtered is evenly dividing along the vertical direction as several regions, the tonal gradation in each region is dropped
It is 16 grades, calculates separately out each region co-occurrence matrix that distance is 1 between vertically downward direction, point pair, and calculates each total
The contrast level parameter of raw matrix；
5) it is supreme down, calculate each piece of region and upper one piece of region the ratio between co-occurrence matrix contrast, wherein being compared
It is worth maximum region as the region where sea horizon；
6) gradient magnitude of region two pieces of region vertical direction adjacent thereto where sea horizon is calculated, with each pixel
The gradient magnitude of point vertical direction calculates the threshold size of this group of data using maximum variance between clusters as initial data,
Pixel using vertical gradient value greater than threshold value is cast out less than the pixel of threshold value and is not considered as candidate point；
7) Hough transform straight-line detection is carried out using obtained candidate point, the best straight line detected is sea horizon.
Further, the colored sea image in the step 1) is 24,3 channel, the colour that maximum brightness grade is 255
Sea image.
Further, in the step 3), gray level image is filtered using by normalized 3 × 3 mask, with
Reduce the calculation amount of filtering operation, filtering concrete operations are successively with mask according to from left to right, and sequence from top to down is right
Each of image pixel is scanned, and the weighted average of pixel goes to replace template center's pixel in the field determined with template
The gray value of point improves signal noise ratio (snr) of image its advantage is that can mitigate the noise jamming in image.
Further, in the step 4), the tonal gradation of filtered image is adjusted to 0~15 by 0~255, then gray scale is total
The matrix that raw matrix is 16 × 16 by 255 × 255 matrix reduction, reduces the calculation amount of follow-up work, greyscale transformation formula
Are as follows:
Wherein, f (x, y) is image initial gray value, and g (x, y) is transformed gray value, and [x] is the maximum no more than x
Integer.
Image uniform is vertically divided into 10 equal parts, successively calculate every parts of images direction vertically downward, point pair
Between distance be 1 gray level co-occurrence matrixes and its contrast level parameter.Assuming that image size is M × N, wherein M is the row of image pixel
Number, N are the columns of image pixel；Coordinate system is established, if the image upper left corner is coordinate origin, positive direction of the x-axis is y vertically downward
Axis positive direction is horizontally to the right, to calculate image vertically, the co-occurrence matrix that distance is 1 between point pair are as follows:
P (i, j)=# (x, y), and (x+1, y) ∈ M × N | f (x, y)=i, f (x+1, y)=j } (2)
Wherein, # (m) is the number of element in set m, and the value range of i, j is 0,1,2 ..., 15, (x, y) and (x+1,
It y) is the position of the pixel in image, f (x, y), f (x+1, y) are the gray value at the pixel.
The contrast calculation formula of gray level co-occurrence matrixes are as follows:
Wherein, the value range of i, j are 0,1,2,3 ..., and 15, P (i, j) are value of the gray level co-occurrence matrixes at (i, j).
Further, in the step 5), the contrast level parameter of every piece of area image is calculated by formula (2) and formula (3)
For con1,con2,,con3…,con10, the ratio between the contrast in the relatively upper block region in every piece of region hiAre as follows:
Wherein, the value range of i is 2,3 ..., 10.Assuming that hmFor maximum value, then m-1, m, the region conduct of m+1 block are taken
The rough estimate position of sea horizon, and guarantee that sea horizon appears in the area image.
Further, in the step 6), the gradient of image vertical direction, which calculates, utilizes first-order difference formula:
Grad (x, y)=f (x+1, y)-f (x, y) (5)
Wherein, f (x+1, y), f (x, y) are respectively the gray value in image at (x+1, y) and (x, y).
Maximum variance between clusters seek the calculating process of optimal threshold are as follows: assuming that according to threshold value Th and gradient value grad (x, y)
Image slices vegetarian refreshments is divided into two classes, wherein gradient value is divided into A class more than or equal to threshold value, and gradient value is divided into B less than threshold value
Class, NAAnd NBThe respectively pixel quantity of A class and B class, then the gradient average value mu of A class and B classAAnd μBAre as follows:
Calculate the inter-class variance of two class of A, B are as follows:
When inter-class variance σ (Th) is maximum, corresponding threshold value Th is optimal threshold, and gradient value is greater than to the point of threshold value
Straight-line detection is carried out as candidate point.
Embodiment:
Below with example come illustrate it is disclosed by the invention it is a kind of based on image divide and textural characteristics sea horizon detection side
Method.The present embodiment realizes that specific implementation step is as follows using C++ programming language and the library OpenCV:
(1) original colored sea image is obtained；
Original sea image is 24,3 channel color image, and resolution ratio is 528 × 938, as shown in Figure 2 a；
(2) original color image is converted into gray level image；
3 channel, 24 color images are converted to 8 gray level images, as shown in Figure 2 b；
(3) gaussian filtering process is carried out to gray level image；
With 3 × 3 masks according to from left to right, sequence from top to down sweeps each of gray level image pixel
It retouches, the weighted average of pixel removes the gray value instead of template center's pixel in the field determined with template, as a result such as Fig. 2 c institute
Show；
(4) filtering image is evenly dividing along the vertical direction the co-occurrence matrix that each region is calculated for several regions and its
Contrast level parameter；
The tonal gradation of image is adjusted to 0~15 by 0~255 using formula (1), is vertically divided into image averaging
The gray level co-occurrence matrixes of every piece of area image and its right are calculated according to formula (2) and formula (3) as shown in Figure 2 d in ten pieces of regions
Than degree parameter；
(5) the ratio between relatively upper one piece of region contrast in every piece of region is calculated, sea horizon rough position is obtained；
Show that contrast changes maximum region according to formula (4), and two pieces of regions adjacent thereto are thick as sea horizon
Slightly estimate region；Fig. 2 e by calculated contrast changes maximum region, Fig. 2 f is sea horizon rough estimate figure；
(6) gradient magnitude for calculating rough region vertical direction calculates threshold size using maximum variance between clusters, selects
Candidate point；
Vertically each pixel gradient magnitude of rough estimate area image is calculated according to formula (5), according to formula (6), formula
(7) and formula (8) calculates optimal threshold, and the pixel that gradient value is greater than threshold value is carried out straight-line detection as candidate point；
(7) straight-line detection is carried out using Hough transform, the best straight line detected is sea horizon, as shown in Figure 2 g.
Claims (5)
1. a kind of sea horizon detection method divided based on image with textural characteristics, which comprises the following steps:
1) original colored sea image is obtained；
2) original color image is converted into gray level image；
3) gaussian filtering process is carried out to gray level image using gaussian filtering frame；
4) image filtered is evenly dividing along the vertical direction as several regions, the tonal gradation in each region is reduced to 16
Grade calculates separately out each region distance between vertically downward direction, point pair and is 1 co-occurrence matrix, and calculates each symbiosis square
The contrast level parameter of battle array；
5) it is supreme down, calculate each piece of region and upper one piece of region the ratio between co-occurrence matrix contrast, wherein obtaining ratio most
Big region is as the region where sea horizon；
6) gradient magnitude for calculating region two pieces of region vertical direction adjacent thereto where sea horizon, is hung down with each pixel
Histogram to gradient magnitude as initial data, the threshold size of this group of data is calculated using maximum variance between clusters, will be hung down
Pixel of the histogram to gradient value greater than threshold value is cast out less than the pixel of threshold value and is not considered as candidate point；
7) Hough transform straight-line detection is carried out using obtained candidate point, the best straight line detected is sea horizon.
2. the sea horizon detection method according to claim 1 divided based on image with textural characteristics, which is characterized in that institute
It states in step 3), gray level image is filtered using by normalized 3 × 3 mask, successively image is pressed with mask
According to from left to right, sequence from top to down is scanned each of image pixel, with pixel in the determining field of template
Weighted average remove the gray value instead of template center's pixel.
3. the sea horizon detection method according to claim 1 divided based on image with textural characteristics, which is characterized in that institute
Its for stating step 4) comprises the concrete steps that:
1. image grayscale grade is adjusted to 0~15 by 0~255, greyscale transformation formula are as follows:
Wherein, f (x, y) is image initial gray value, and g (x, y) is transformed gray value, and [x] is whole no more than the maximum of x
Number；
2. image uniform is divided into 10 equal parts along the vertical direction, successively calculate every parts of images direction vertically downward, point pair between
Gray level co-occurrence matrixes and its contrast level parameter of the distance for 1；Assuming that image size is M × N, wherein M is the line number of image pixel,
N is the columns of image pixel, establishes coordinate system, if positive direction of the x-axis is that vertically downward, positive direction of the y-axis is that horizontally to the right, image is left
Upper angle is coordinate origin, calculates image vertically, the co-occurrence matrix that distance is 1 between point pair are as follows:
P (i, j)=# (x, y), and (x+1, y) ∈ M × N | f (x, y)=i, f (x+1, y)=j } (2)
Wherein, # (m) is the number of element in set m, and the value range of i, j are 0,1,2 ..., 15, (x, y) and (x+1, y) is
The position of pixel in image, f (x, y), f (x+1, y) are the gray value at the pixel；
3. calculating the contrast C on of gray level co-occurrence matrixes:
Wherein, the value range of i, j are 0,1,2,3 ..., and 15, P (i, j) are value of the gray level co-occurrence matrixes at (i, j).
4. the sea horizon detection method according to claim 1 divided based on image with textural characteristics, which is characterized in that institute
It states in step 5), the gray level co-occurrence matrixes and its contrast level parameter in every piece of region is obtained by calculation, contrast is changed maximum
Rough estimate position as sea horizon of region and its two pieces of adjacent regions, the contrast value in every piece of region is coni, i's
Value range is 2,3 ..., 10, the ratio between the contrast in the relatively upper block region in every piece of region hiAre as follows:
Assuming that hmFor maximum value, then m-1, m, rough estimate position of the m+1 block region as sea horizon are taken.
5. the sea horizon detection method according to claim 1 divided based on image with textural characteristics, which is characterized in that institute
State comprising the concrete steps that for step 6):
1. calculating the gradient of image vertical direction using first-order difference formula:
Grad (x, y)=f (x+1, y)-f (x, y) (5)
Wherein, f (x+1, y), f (x, y) are respectively the gray value in image at (x+1, y) and (x, y)；
2. utilizing maximum between-cluster variance algorithm threshold value, gradient is divided into two class of A, B, then inter-class variance are as follows:
Wherein, NAAnd NBThe respectively pixel quantity of A class and B class, μAAnd μBFor the gradient average value of A class and B class, μAAnd μBMeter
Calculate formula are as follows:
According to formula (6), formula (7) and formula (8), when σ (Th) is maximum, corresponding threshold value Th is optimal threshold；
3. the pixel that gradient is greater than or equal to threshold value is classified as candidate point by the threshold value according to obtained in 2., gradient value is less than threshold
The pixel of value, which is cast out, not to be considered.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810768241.9A CN109242870A (en) | 2018-07-13 | 2018-07-13 | A kind of sea horizon detection method divided based on image with textural characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810768241.9A CN109242870A (en) | 2018-07-13 | 2018-07-13 | A kind of sea horizon detection method divided based on image with textural characteristics |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109242870A true CN109242870A (en) | 2019-01-18 |
Family
ID=65072683
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810768241.9A Pending CN109242870A (en) | 2018-07-13 | 2018-07-13 | A kind of sea horizon detection method divided based on image with textural characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109242870A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109886899A (en) * | 2019-03-12 | 2019-06-14 | 哈尔滨工程大学 | A kind of sea horizon image detecting method by dynamic disturbance |
| CN109978869A (en) * | 2019-03-29 | 2019-07-05 | 清华大学 | A kind of sea horizon detection method and system based on gray level co-occurrence matrixes and Hough transform |
| CN110276788A (en) * | 2019-06-12 | 2019-09-24 | 北京轩宇空间科技有限公司 | Method and apparatus for infrared imaging formula target seeker target following |
| CN110298855A (en) * | 2019-06-17 | 2019-10-01 | 上海大学 | A kind of sea horizon detection method based on gauss hybrid models and texture analysis |
| CN110619622A (en) * | 2019-04-08 | 2019-12-27 | 天津职业技术师范大学(中国职业培训指导教师进修中心) | Bread board structure image automatic detection method based on computer vision |
| CN111027546A (en) * | 2019-12-05 | 2020-04-17 | 北京嘉楠捷思信息技术有限公司 | Character segmentation method and device and computer readable storage medium |
| CN111047570A (en) * | 2019-12-10 | 2020-04-21 | 西安中科星图空间数据技术有限公司 | Automatic cloud detection method based on texture analysis method |
| CN112150480A (en) * | 2020-09-17 | 2020-12-29 | 哈尔滨工业大学（威海） | Background segmentation method based on sea-sky-line |
| CN112700441A (en) * | 2021-01-28 | 2021-04-23 | 中北大学 | Automatic weld searching method based on textural features |
| CN113705501A (en) * | 2021-09-02 | 2021-11-26 | 浙江索思科技有限公司 | Offshore target detection method and system based on image recognition technology |
| CN114219794A (en) * | 2021-12-17 | 2022-03-22 | 沭阳县桐盛木业制品厂(普通合伙) | Method and system for evaluating surface quality of shaving board based on machine vision |
| CN115147733A (en) * | 2022-09-05 | 2022-10-04 | 山东东盛澜渔业有限公司 | Artificial intelligence-based marine garbage recognition and recovery method |
| CN116843689A (en) * | 2023-09-01 | 2023-10-03 | 山东众成菌业股份有限公司 | Method for detecting surface damage of fungus cover |
| CN113705501B (en) * | 2021-09-02 | 2024-04-26 | 浙江索思科技有限公司 | Marine target detection method and system based on image recognition technology |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103198455A (en) * | 2013-03-14 | 2013-07-10 | 哈尔滨工程大学 | Image denoising method utilizing total variation minimization and gray scale co-occurrence matrixes |
| CN105184804A (en) * | 2015-10-12 | 2015-12-23 | 上海海大技术转移有限公司 | Sea surface small target detection method based on airborne infrared camera aerially-photographed image |
| CN106530313A (en) * | 2016-11-16 | 2017-03-22 | 上海大学 | Sea-sky line real-time detection method based on region segmentation |
-
2018
- 2018-07-13 CN CN201810768241.9A patent/CN109242870A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103198455A (en) * | 2013-03-14 | 2013-07-10 | 哈尔滨工程大学 | Image denoising method utilizing total variation minimization and gray scale co-occurrence matrixes |
| CN105184804A (en) * | 2015-10-12 | 2015-12-23 | 上海海大技术转移有限公司 | Sea surface small target detection method based on airborne infrared camera aerially-photographed image |
| CN106530313A (en) * | 2016-11-16 | 2017-03-22 | 上海大学 | Sea-sky line real-time detection method based on region segmentation |
Non-Patent Citations (2)
| Title |
|---|
| DONG LIANG 等: "Robust sea-sky-line detection for complex sea background", 《2015 IEEE INTERNATIONAL CONFERENCE ON PROGRESS IN INFORMATICS AND COMPUTING (PIC)》 * |
| 戴永寿 等: "基于局部Otsu分割与Hough变换的海天线检测", 《光电工程》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109886899A (en) * | 2019-03-12 | 2019-06-14 | 哈尔滨工程大学 | A kind of sea horizon image detecting method by dynamic disturbance |
| CN109978869A (en) * | 2019-03-29 | 2019-07-05 | 清华大学 | A kind of sea horizon detection method and system based on gray level co-occurrence matrixes and Hough transform |
| CN110619622A (en) * | 2019-04-08 | 2019-12-27 | 天津职业技术师范大学(中国职业培训指导教师进修中心) | Bread board structure image automatic detection method based on computer vision |
| CN110619622B (en) * | 2019-04-08 | 2023-07-14 | 天津职业技术师范大学(中国职业培训指导教师进修中心) | Bread board structure image automatic detection method based on computer vision |
| CN110276788A (en) * | 2019-06-12 | 2019-09-24 | 北京轩宇空间科技有限公司 | Method and apparatus for infrared imaging formula target seeker target following |
| CN110298855B (en) * | 2019-06-17 | 2023-05-16 | 上海大学 | Sea-sky-line detection method based on Gaussian mixture model and texture analysis |
| CN110298855A (en) * | 2019-06-17 | 2019-10-01 | 上海大学 | A kind of sea horizon detection method based on gauss hybrid models and texture analysis |
| CN111027546A (en) * | 2019-12-05 | 2020-04-17 | 北京嘉楠捷思信息技术有限公司 | Character segmentation method and device and computer readable storage medium |
| CN111047570B (en) * | 2019-12-10 | 2023-06-27 | 中科星图空间技术有限公司 | Automatic cloud detection method based on texture analysis method |
| CN111047570A (en) * | 2019-12-10 | 2020-04-21 | 西安中科星图空间数据技术有限公司 | Automatic cloud detection method based on texture analysis method |
| CN112150480A (en) * | 2020-09-17 | 2020-12-29 | 哈尔滨工业大学（威海） | Background segmentation method based on sea-sky-line |
| CN112700441A (en) * | 2021-01-28 | 2021-04-23 | 中北大学 | Automatic weld searching method based on textural features |
| CN113705501A (en) * | 2021-09-02 | 2021-11-26 | 浙江索思科技有限公司 | Offshore target detection method and system based on image recognition technology |
| CN113705501B (en) * | 2021-09-02 | 2024-04-26 | 浙江索思科技有限公司 | Marine target detection method and system based on image recognition technology |
| CN114022473B (en) * | 2021-11-19 | 2024-04-26 | 中国科学院长春光学精密机械与物理研究所 | Horizon detection method based on infrared image |
| CN114219794A (en) * | 2021-12-17 | 2022-03-22 | 沭阳县桐盛木业制品厂(普通合伙) | Method and system for evaluating surface quality of shaving board based on machine vision |
| CN114219794B (en) * | 2021-12-17 | 2023-01-20 | 沭阳县桐盛木业制品厂(普通合伙) | Method and system for evaluating surface quality of shaving board based on machine vision |
| CN115147733A (en) * | 2022-09-05 | 2022-10-04 | 山东东盛澜渔业有限公司 | Artificial intelligence-based marine garbage recognition and recovery method |
| CN115147733B (en) * | 2022-09-05 | 2022-11-25 | 山东东盛澜渔业有限公司 | Artificial intelligence-based marine garbage recognition and recovery method |
| CN116843689A (en) * | 2023-09-01 | 2023-10-03 | 山东众成菌业股份有限公司 | Method for detecting surface damage of fungus cover |
| CN116843689B (en) * | 2023-09-01 | 2023-11-21 | 山东众成菌业股份有限公司 | Method for detecting surface damage of fungus cover |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109242870A (en) | A kind of sea horizon detection method divided based on image with textural characteristics | |
| CN111145161B (en) | Pavement crack digital image processing and identifying method | |
| CN111310558B (en) | Intelligent pavement disease extraction method based on deep learning and image processing method | |
| CN108805904B (en) | Moving ship detection and tracking method based on satellite sequence image | |
| CN106557764B (en) | A kind of water level recognition methods based on binary-coded character water gauge and image procossing | |
| CN109187581A (en) | The bearing finished products plate defects detection method of view-based access control model | |
| CN105868745B (en) | Weather recognition methods based on dynamic scene perception | |
| CN107610114A (en) | Optical satellite remote sensing image cloud snow mist detection method based on SVMs | |
| CN108846844B (en) | Sea surface target detection method based on sea antenna | |
| CN107292879B (en) | A kind of sheet metal surface method for detecting abnormality based on image analysis | |
| CN110210448B (en) | Intelligent face skin aging degree identification and evaluation method | |
| CN111161222B (en) | Printing roller defect detection method based on visual saliency | |
| CN112233116B (en) | Concave-convex mark visual detection method based on neighborhood decision and gray level co-occurrence matrix description | |
| CN108470338B (en) | A kind of water level monitoring method | |
| CN109376740A (en) | A kind of water gauge reading detection method based on video | |
| CN115294159B (en) | Method for dividing corroded area of metal fastener | |
| CN109975196B (en) | Reticulocyte detection method and system | |
| CN111060442A (en) | Oil particle detection method based on image processing | |
| CN115100174B (en) | Ship sheet metal part paint surface defect detection method | |
| CN110807763A (en) | Method and system for detecting ceramic tile surface bulge | |
| CN107463939B (en) | Image key straight line detection method | |
| CN109858484B (en) | Multi-class transformation license plate correction method based on deflection evaluation | |
| CN109961065B (en) | Sea surface ship target detection method | |
| Lin et al. | License plate location method based on edge detection and mathematical morphology | |
| CN110930425B (en) | Damaged target detection method based on neighborhood vector inner product local contrast image 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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190118 |